Skip to main content

Colour Blindness Simulator

How your colour pair reads to everyone else.

Two colours to compare

Typically two you are relying on to be told apart — chart series, a status indicator, a before and after.

How the pair reads

  • Typical colour vision

  • Protanopia~1% of men

    Red cones absent. Reds darken heavily and red/green separation collapses.

    Distinguishable

    #62592D · #918234 — separation 0.145

  • Deuteranopia~1% of men

    Green cones absent. The most common form — red and green become near-identical.

    Too similar

    #8B7C28 · #867A42 — separation 0.028

  • Tritanopia~0.01%

    Blue cones absent. Blue and yellow confuse; rare and not inherited by sex.

    Distinguishable

    #E80032 · #288A7C — separation 0.325

  • Achromatopsia~0.003%

    No colour vision at all. Also a useful proxy for greyscale printing.

    Too similar

    #707070 · #7D7D7D — separation 0.046

Separation is a perceptual distance measured in OKLab after the deficiency is applied. Under about 0.05 the two are effectively one colour. This is a different question from contrast: a pair can pass every WCAG contrast check against the background and still be indistinguishable from each other, which is why colour must never be the only way information is conveyed.

Compare two colours as they appear under each of the four colour vision deficiencies, with a numeric score for how far apart they remain. This is the check a contrast tool cannot do: two colours can both pass contrast against the background and still be indistinguishable from each other, which is how a chart legend fails for one man in twelve while passing every automated audit.

01

Why use this tool?

Four deficiencies

Protanopia, deuteranopia, tritanopia and achromatopsia, with how common each is.

A number, not a vibe

Perceptual separation measured in OKLab after the deficiency is applied. Under about 0.05 the pair is effectively one colour.

Accurate transforms

Machado et al. matrices applied in linear RGB, which is the step most implementations skip.

Catches what contrast misses

A pair can pass every WCAG check and still be unusable together.

02

How this colour blindness simulator works

The simulation uses the Machado, Oliveira and Fernandes transforms at full severity — the matrices most accessibility tooling is built on. They are applied in linear RGB rather than to the gamma-encoded values directly. That step is the one quick implementations skip, and skipping it makes results too dark and too saturated, which understates how much of a design actually collapses.

Separation is measured as a perceptual distance in OKLab after the deficiency is applied, which turns a subjective judgement into a number. Below roughly 0.05 the two colours are effectively one. Judging simulated swatches by eye is unreliable for someone with typical colour vision, who has no intuition for what has been removed.

Deuteranopia and protanopia — the two red-green types — account for the overwhelming majority of cases and affect around one man in twelve. Tritanopia and achromatopsia are far rarer, but achromatopsia doubles as a useful proxy for greyscale printing and for anyone reading a monochrome screenshot.

03

How to use it

  1. Step 1: Enter two colours

    Usually two you need people to tell apart — chart series, status indicators.

  2. Step 2: Read down the list

    Each row shows the pair as it appears under that deficiency.

  3. Step 3: Watch the separation score

    Below 0.05 you need a second signal — a pattern, a label, a different shape.

04

Example usage

Chart series
The classic red-and-green pair, which passes every contrast check against white and becomes a single colour under deuteranopia.
Status indicators
Success and error states relying on colour alone, tested before someone cannot tell which one they are looking at.
Map keys
Adjacent regions in similar hues, where a small separation on screen becomes none at all for some readers.
05

What a simulation can and cannot tell you

The transforms on this page model dichromacy — the complete absence of one cone type. That is the severe end of the range, and it is the right thing to design against, but it is not what most people with a colour vision deficiency actually experience. Anomalous trichromacy, where a cone type is present but shifted, is considerably more common and produces a milder and much more variable effect.

This matters for how you read the results. A pair that survives full deuteranopia will work for essentially everyone, so a pass here is genuinely reassuring. A pair that fails is not necessarily invisible to every affected reader — but it is unreliable, and unreliable is enough reason to fix it. Simulating at full severity turns a spectrum of individual experience into one conservative test you can actually design against.

What no simulation conveys is the lived part. Someone with a lifelong deficiency has spent decades developing compensations — reading position, brightness, context, learned convention — that a person seeing a simulated image for the first time has none of. Treat these swatches as an engineering check on whether information survives, not as a window into anyone's experience.

06

Frequently asked questions

How common is colour blindness?

Around one in twelve men and one in two hundred women have some form. Deuteranopia and protanopia — the red-green types — account for the overwhelming majority.

Why do you show a number as well as a picture?

Because judging two simulated swatches by eye is unreliable, particularly for someone with typical colour vision who has no intuition for what has been lost. The number is a measurement.

What should I do if a pair fails?

Add a second channel. Different line styles, a pattern fill, direct labels, or different shapes. The rule is that colour should never be the only thing carrying information — which also helps in greyscale printing.

Is this what colour blindness actually looks like?

Not exactly. These are dichromacy transforms, modelling the complete absence of a cone type. Anomalous trichromacy — a shifted rather than missing cone — is more common and milder, and it varies considerably between individuals. Simulating the severe case gives you one conservative test rather than a portrait of anyone's vision.

Which colour pairs are reliably safe?

Pairs separated by lightness rather than hue survive almost everything, since lightness is the channel every deficiency preserves. Blue against orange is the classic safe hue pair, because the blue-yellow axis is untouched by the common red-green types. Red against green is the classic unsafe one.

Does dark mode change any of this?

It changes the lightness relationships, so a pair verified on white needs rechecking on a dark surface. Saturated colours also tend to read as more intense against dark backgrounds, which can reduce the apparent separation between two of them. Test the pairs in both themes rather than assuming one result transfers.

07

Browse every tool in Colour Tools.